Weather Forecasting

- The system is going to estimate the main weather parameters (GHI, GTI, ambient
temperature, wind speed and direction) from the next hours / up to 8 days with
medium and long-term forecasts with meteorological models for an efficient grid
integration. using different functionality to provide:
- Current weather conditions
- Now-casting (every minute to monitor cloud movement and GHI/GTI till 30
minutes)
- Short-term forecast (every 30 minutes till following 2 hours minimum).
- Medium-term forecast and long-term forecast (from next hours to 8 days).
- Weather forecast data to be provided every 5 minutes for the following 72 hours.
- Cloudiness impact on the solar field.
- Comparison between weather forecast and real weather.
- The weather forecast will feed the generation forecast system with several
meteorological variables to support the generation forecast models and
calculation
- Output file to be generated as per format of performance model.
The output file of the WFS shall meet the requirements of the Generation Weather (GFS)
in terms of communication with the GFS. Additionally, weather forecast report to be
submitted via email to a predefined list of recipients in excel format and to be agreed as
per Grid Operator requirements
Weather data shall be available daily before 6AM (Local Time) with the following 72 hours
and updated with the intraday forecast every 3 hours.
The GFS system is mainly used to forecast the power generation of PV plants. Based on
high-precision numerical weather forecasts, a complete database system is built, and
various communication interfaces are used to collect measured weather data, numerical
weather data, and actual power data.
After the weather server receives the weather forecast data and processes it, it transmits
the data to the generation forecast server through the reverse isolator. The generation
forecast server is connected to the station and the SCADA system through the firewall, to
collect, store, count and analyse the actual power.
- 5 min basis report with the net electricity generation for the following 72 hours
- 30 min basis report with the net electricity generation for following 2 hours
- 1 hour basis report with the net electricity generation forecast for next day up to 8
days.
The generation forecast server performs parallel calculation and processing based on the
received weather data, real-time data, and inverter data, and can obtain a 192-hour
medium-long term generation forecast, and a 2hour short-term generation forecast curve
in the future. According to IEC 61850, the generation forecast system is separately
equipped with a communication management machine to communicate with SCADA.
Generation forecast should be available daily before 8AM (Local Time) with the following
72 hours and updated with the intraday forecast every 3 hours. It should be
communicated with DCS as per required format, and, additionally, to be submitted via
email to a predefined list of recipients in excel format and to be agreed as per Grid
Operator requirements.
Manual simulations could be launched in case of unexpected events, such as equipment
unavailability, in the intraday period.

Requirements
The scope of supply shall meet the following requirements:
- Supply the hardware and software to properly functioning of the system.
This includes:
o On-site sensors and cameras (if need)
o numerical weather prediction models and,
o external weather information such as weather satellites, for predicting future general
weather trends 8 days ahead. The satellite data will be available on-line from other
sources.
- The final data to be provided by this system in the control room to plant operators will
be following different formats:
o Several images and weather forecast for several meteorological variables (i.e. GHI, GTI,
ambient temperature, Wind speed and direction, etc). Some of these images will be
provided locally such as the local sky images and other would be accessible through a
secure server on the internet, such as satellite images.
o the information will be available on a weather forecast computer screen/ computer
screen in the control room and delivered by communications.
- Providing the utilities forecast system the information described in this document.
Services:
- Technical assistance during hardware installation.
- Commissioning of WFS.
- Training courses for control room operators and operation manager.
- Monthly Weather data accuracy report.
- Daily comparison of the values below with morning forecast for the next day.
- GHI, GTI measured by plant meteorological stations vs GHI and GTI estimated by WFS.
- Maintenance of the system including spares (defined by the Client) and failed elements.
o Hardware maintenance by others (supplier shall get advice)
o Software maintenance by Supplier
– the Supplier shall provide a quotation for maintenance services)
- Integration, communication & Testing with weather station and DCS,.., etc
Hardware:
- All the components (with all their accessories to ensure a correct operation) included in
this specification shall be supplied.
Each system includes the following equipment:
o One (1) PC Workstation (of a remarkable brand)
o Weather data processing server
o Network security equipment: security isolation devices, firewalls, etc.
Software:
- Software package for WFS and GFS including its installation:
o Real-time and historical data processing
o Nowcasting
o Satellite imagery
o Numerical Weather Prediction (NWP)
- Subscription agreement for a year with the weather forecast company that provides
satellite imagery, and a vendor list of three (3) possible suppliers.
Power supply with the following characteristics will be available for equipment and
accessories:
- 400 Vac, three phase + PE, 50 Hz
- 230 Vac, single-phase + neutral + PE, 50 Hz
The remaining voltages required shall be obtained by the Supplier from the above
available power supplies with the necessary power conversion system.
The relevant equipment is provided by the Supplier. The Plant components shall be
designed, manufactured, and tested at the factory to the maximum extent possible. The
Supplier shall provide a detailed supply list, which in turn describes the model, quantity,
origin, manufacturer and so on. For parts that are required for the operation and
construction of the entire equipment, even if this annex is not listed or the number is
insufficient, the supplier shall make up the contract at the time of execution without any
cost.
The WFS and GFS will provide site specific weather and forecasts in two ways:
- Real time estimation.
- Forecast for next 2 hours.
- Forecast for next 8 days.
Combining information of three (3) different methods:
Nowcasting, satellite imagery and NWP, the WFS shall provide reports with information
below (regarding the two ways indicated above):
- Time:
o Date (day, month, year)
o Standard Time
o Solar Time
o GHI
o GTI
o Wind direction and wind speed
o Dry Bulb Temperature
Now-casting
Now-casting method uses cameras installed at site. Supplier shall justify the use of more
cameras if needed and support with the guidelines to select the suitable location of
them. Now-casting software will take information from Meteorologic Stations and pictures
from sky cameras every minute to foresee the weather in a range of 1 to the next 30
minutes.
The Now-casting shall:
- Provide a projection of the cloud’s shadow overlapped with the layout of the solar field
every minute.
- Provide a projection of the expected cloud’s shadow overlapped with the layout of the
solar field every minute during next 30minutes.
- Show the real images in a web service platform.
- Record last 15 minutes while visualization in video mode is running.
- Visualization in full-screen.
- Record Historical images of the last 30 days (minimum) for searching and analysing as
maximum as possible.
The Supplier may propose an alternative architectural control for the camera system
upon Client or its representative’s approval.
Satellite Imagery
Satellite imagery can provide a forecast when a medium-term response WFS software is
applied over collected satellite pictures from a Geographical Information System (GIS)
platform. Using two (2) successive satellite images is possible to obtain an estimation of
the clearness index for every pixel using a radiative transfer model for clear sky
situations. With this evaluation using the visible channel, plus geometrical, statistical and
stocastical methods to calculate the corresponding direct normal irradiance (DNI), global
normal irradiance (GNI), photosynthetically active radiation (PAR) or other integrated
(non-spectral) components, the software will provide the global horizontal irradiance
(GHI). The satellite imagery shall provide:
GHI and GTI forecasts provided as a time series. This information will be saved in CSV
format (to be defined later) and will be communicated by FTP or similar (to be defined
later).
- The forecast time range will be 2 hours, and data series will be updated every 15
minutes. 1 minute granularity will be considered.
- Real cloudiness satellite images will be shown on an intranet based on a Geographical
Information System (GIS) platform. Intranet will be user friendly including types of zoom,
image size and display of the last hours among others.
- Graphical data visualization with the forecast information of GHI and GTI
- Real time cloudiness information for the last 3 hours in a zoomable platform.
- Prediction of cloud movement and direction in the satellite image. Distance and speed
will be indicated.
- Prediction of dust and precipitation.
- Visualization in full- screen will be provided.
- Historical visualization of satellite images of minimum the last 30 days will be available
for searching and analyzing.
Regarding software supply:
- System shall collect satellite pictures from a GIS platform.
Numerical Weather prediction
Mesoscale numerical weather prediction (NWP) model forecast provides from short to
long-term predictions based on geophysical circulation equations which take into account
the different interactions and dynamics of the atmosphere-ocean-land system at
mesoscale level (few tens of km), so higher resolution is achieved, and smaller scale
atmospheric phenomena could be treated. The system of physical equations applied to
different nesting is solved by numerical methods that require initial and boundary
conditions of the atmospheric state. The generation of these conditions is obtained from
previous simulations with NWP that operate on a global scale and are easily loaded and
assimilated by the model. A site-adapted model is usually prior developed in order to
improve the forecast accuracy using historical data series and comparing the outputs
obtained with real data series of the same period to extract several statistics values that
give information of the forecast deviation.
NWP service shall provide:
- NWP forecast computation takes place at external servers and shall be accessible by
web application. Forecasts will be saved in CSV (to be defined later) and will be
communicated by FTP or similar (to be defined later).
- The forecast time range will be 8 days, and it will be updated every 1 hours.
Meteorological data file with forecast will have 15 minutes granularity.
- Meteorological variable forecasts will be shown in maps with the position of the plant
and in meteograms with numerical data. Alarms or high wind speed, precipitation and
dust will be included.
The Generation forecasting system adopts mature products that have been successfully
applied to grid connected photovoltaic power plants. The generation forecasting system
equipment shall be able to predict the short-term and ultra short-term photovoltaic
power, and the prediction is relatively accurate.
The system shall have data processing, generation forecasting, interface query statistics
and other functions. The Supplier shall put in place a forecasting system which is
compliant with the most recent version of the Renewables Standards.
The GFS adopts mature products (based on AI modelling) that have been successfully
applied to grid connected photovoltaic power plants. The power forecasting system
equipment shall be able to predict the medium/long-term forecast and short-term
photovoltaic power, and the prediction is relatively accurate. The system shall have data
processing, power forecasting, interface query statistics and other functions. The
Supplier shall put in place a forecasting system which is compliant with the most recent
version of the Renewables Standards. In absence of a clear function or detailed
description of the forecasting requirements in the codes and standards.
The Supplier shall propose a suitable methodology for addressing the requirements. This
method statement shall include, but not be limited to:
• Identification and selection of the appropriate model (referred to in the table above as
the “Model”), being either a forecast based on a Numerical Weather Prediction model, a
Multi Model Forecast or any suitable combination thereof;
• Generation forecast include should GHI, GTI, MWh, MVAR (reactive forecasted power).
• Identification and selection of the appropriate input data sets, including:
• Weather model information.
• Satellite imagery sources along with calculation methods and quality checks
• Local Plant sensors along with calculation methods and quality checks.
• Description of post processing methods for the model outputs utilised by the forecast
provider;
• List of information and parameters provide with each forecast in compliance with the
requirements of the norms, code and regulations.
• Proposal of method of integration of forecast information in the Plant DCS (local
SCADA) at the O&M building / plant control centre;
• There must be provision to send all generation forecast data (daily, every 30 min, 2
hourly, real time etc. as per Offtaker require and it should be configurable) automatically
to Offtaker through DCS system.
• DCS SCADA system shall have one screen or provision to show the requested data with
graphical view.
• Provision of the required information in the format and under the data protocol as well
as data exchange methods required by the Offtaker.
• The generation forecast system data should be comply with Offtaker requested format
data.
• Data storage function in DCS must be provided.
• Assurance of compliance with Cyber Security Requirements in data exchange with
external providers, the Offtaker and internal information processing.
• Description of testing, proving of functional compliance and required performance of
the parameters as well as trial operating and auditing. Based on the method statement
the details of the forecasting, including accuracy, will be defined between the Generator
and the Offtaker before project implementation.
In addition to the integration into the Plant DCS and the provision of the forecast
parameters required in the norms, code and regulations to the Offtaker, the Generator
and the Offtaker shall have access to the online portal of the weather forecast provider
where additional information on a different forecast model, different time horizons,
deltas, ramp warnings or uncertainties shall be made available.
The data needed for the operation of the power forecasting system includes the
numerical weather prediction data (the numerical weather prediction data comes from
the domestic authoritative meteorological department), the automatic environmental
monitoring station data, the real-time output power data of the PV plant, the operation
status of the photovoltaic array, etc. The system shall be able to complete the automatic
collection of all data, and can also be entered manually, in which the numerical weather
forecast data will be acquired automatically and regularly.
Data statistics function
(1) The time range for the system to participate in the statistics can be arbitrarily
selected.
(2) Historical power data statistics, historical meteorological data statistics, etc.
(3) The system shall count the errors between the predicted results and the actual power
generation of various parameters under different conditions. The statistical results can be
created and exported dynamically for the use of subsequent analysis and scientific
research.
(4) The system shall be able to carry out error statistics for the prediction results in any
time interval, and the error indexes shall include root mean square error rate, average
absolute error rate, correlation coefficient, etc.
Data analysis and processing function
(1) The prediction system should have a series of filtering and processing functions for
basic data to reduce the impact of outliers and error information on prediction results.
(2) The prediction system shall be able to interpret and compare the numerical weather
forecast value and the automatic environmental monitoring station data.
(3) The error of medium-term and short-term power forecasting should be calculated by
the absolute average percentage error, that is, the difference between the actual power
generation and the power generation should be included in the error in the form of
absolute value.
(4) The prediction system should realize parameter interaction and analyse the
correlation between different parameters.